OpenAI o3-mini

What o3-mini is

o3-mini is a small reasoning model that OpenAI released on 31 January 2025. It was made available to all ChatGPT users, including people on the free tier, and to some users of the API ^[1]. A reasoning model is one trained to spend extra computation working a problem through before it answers, rather than producing a reply in a single pass; calling it "small" places it in a cheaper, faster tier than the largest models a lab offers.

Effort levels and access

The model offers three reasoning-effort levels — low, medium and high — and the free version runs at medium ^[1]. Paying subscribers could use a higher-compute version, o3-mini-high, and Pro subscribers had unlimited access to both ^[1]. On 12 February 2025 the o3-mini-high allowance for Plus subscribers rose from 50 requests per week to 50 requests per day ^[1]. The point of the tiers is to let a user trade waiting time, and compute cost, against answer quality on the same model.

What developers got

According to OpenAI, o3-mini was the company's first small reasoning model to support function calling, Structured Outputs and developer messages, and it supported streaming ^[4]. Function calling lets the model ask the surrounding program to run a tool and hand back the result, so it can consult a database or a calculator instead of guessing. Structured Outputs constrains a reply to a supplied schema, such as a JSON shape, so that code can parse it without repair work. Developer messages are instructions written by whoever builds the application, kept separate from what the end user types. Streaming means tokens arrive as they are produced rather than all at once at the end. For teams running production systems, features like these often decide whether a model can be wired into a pipeline at all.

Strengths and limits

The model was optimised for STEM reasoning, with particular strength in science, mathematics and coding ^[4]. It did not accept images, and OpenAI pointed developers who needed visual reasoning to o1 instead ^[4].

Price and lifetime

Through the API, o3-mini was priced at $1.10 per million input tokens and $4.40 per million output tokens, with a 200K-token context window ^[2]. A token is the unit a model reads and writes in — roughly a short word or a piece of one — and the context window is the total number of tokens a single request can carry, including instructions, the conversation and the answer. Input and output are billed at different rates. OpenAI's API documentation listed the o3-mini model for shutdown on 23 October 2026 ^[3].

Where it sits

o3-mini was previewed alongside o3 on 20 December 2024, and it was succeeded by o4-mini in April 2025 ^[1]. Its release put that generation's reasoning on the free tier of ChatGPT as well as at a lower API price point ^[1].